Robust bus-stop identification and denoising methodology

The analysis of public transportation data is receiving an increasing amount of attention from the research community in the past few years. This interest is fueled by the widespread installation and open access to a variety of sensor technologies for collecting data on the state of the transport system in many cities around the world. Different cities provide different data sources and in many cases the only common dataset is represented by GPS data of the vehicle fleet. Very often, the data contain erroneous or missing information that should be corrected before proceeding with their analysis. In this paper, we propose a methodology to de-noise scheduled bus stops and detect time schedule information using GPS AVL (Automatic Vehicle Location) data. The methodology performs different sequential steps: i) cleaning process and detection of trips; ii) bus stop extraction; ii) bus stop clustering; iv) feature extraction; v) classification model construction and application. Moreover, the impact on the whole process of different methods applied in different steps is empirically evaluated on datasets with different temporal extent.